Inspiration

I would have liked to learn stick animation for my baby niece then I thought of a simple idea and used multi agentic system to do that.

What it does

Basically it takes a prompt and generates relevant stick figures the agentic flow is

Director -> Researcher -> Animator <-> critic -> User Input if needed

How we built it

Built using google ADK, antigravity, gemini-flash-latest and deployed on google cloud run.

Challenges we ran into

I would say I am worried about the consumption/gcp costs so I haven't shared it with friends or family yet. Rest was fine.

Accomplishments that we're proud of

Coming Up with a way to store and generate using json. It's a simple idea but it uses multimodal capabilities and not something like VEO

What we learned

That there are ways to build things without increasing complexity.

What's next for Dreams Animator

I would say I would like to find a way to use blender and move it towards 3d generation in the same way and generate proper animations.

Built With

  • antigravity
  • cloudrun
  • googleadk
  • python
Share this project:

Updates